Binocular perception of slant about oblique axes relative to a visual frame of reference

Abstract

From the literature it is known that the processing of disparity for slant is different in the presence and in the absence of a visual frame of reference. We elaborate the experimental finding that vertical disparity is not processed for slant perception in the presence of a visual reference. This theoretical analysis results in a reduction of the thre basic first-order transformations between the retinal half-images (Koen derink's divergence, rotation and deformation) to only two basic orthogonal transformations. The fire of htese, horisontal scale, results in slant perception about the vertical axis, whereas the second, horisontal shear, results in slant perception about the horisontal axis. These transformations are based primarily on horisontal disparity. We show experimentally that in the presence of a frame of reference tha amount of vertical transformation that is added to the two basic transformations (horisontal scale and shear) of a random-dot stimulus is indeed irrelevant for slant perception. We suggest that in the presence of a visual reference slant perception about oblique axes is based solely on linear combinations of the horisontal scale and horisontal shear trensformations. Subjects are able to reproduce slants about oblique axes experimentally merely by combining horisontal scale and shear.